Go to the documentation of this file.
1 /*
2  * This file is part of FFmpeg.
3  *
4  * FFmpeg is free software; you can redistribute it and/or
5  * modify it under the terms of the GNU Lesser General Public
6  * License as published by the Free Software Foundation; either
7  * version 2.1 of the License, or (at your option) any later version.
8  *
9  * FFmpeg is distributed in the hope that it will be useful,
10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
12  * Lesser General Public License for more details.
13  *
14  * You should have received a copy of the GNU Lesser General Public
15  * License along with FFmpeg; if not, write to the Free Software
16  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17  */
19 /**
20  * @file
21  * video scene change detection filter
22  */
24 #include "libavutil/avassert.h"
25 #include "libavutil/imgutils.h"
26 #include "libavutil/opt.h"
27 #include "libavutil/pixdesc.h"
28 #include "libavutil/timestamp.h"
30 #include "avfilter.h"
31 #include "filters.h"
32 #include "scene_sad.h"
34 typedef struct SCDetContext {
35  const AVClass *class;
37  ptrdiff_t width[4];
38  ptrdiff_t height[4];
39  int nb_planes;
40  int bitdepth;
42  double prev_mafd;
43  double scene_score;
45  double threshold;
46  int sc_pass;
47 } SCDetContext;
49 #define OFFSET(x) offsetof(SCDetContext, x)
53 static const AVOption scdet_options[] = {
54  { "threshold", "set scene change detect threshold", OFFSET(threshold), AV_OPT_TYPE_DOUBLE, {.dbl = 10.}, 0, 100., V|F },
55  { "t", "set scene change detect threshold", OFFSET(threshold), AV_OPT_TYPE_DOUBLE, {.dbl = 10.}, 0, 100., V|F },
56  { "sc_pass", "Set the flag to pass scene change frames", OFFSET(sc_pass), AV_OPT_TYPE_BOOL, {.dbl = 0 }, 0, 1, V|F },
57  { "s", "Set the flag to pass scene change frames", OFFSET(sc_pass), AV_OPT_TYPE_BOOL, {.dbl = 0 }, 0, 1, V|F },
58  {NULL}
59 };
64 {
65  static const enum AVPixelFormat pix_fmts[] = {
76  };
78  AVFilterFormats *fmts_list = ff_make_format_list(pix_fmts);
79  if (!fmts_list)
80  return AVERROR(ENOMEM);
81  return ff_set_common_formats(ctx, fmts_list);
82 }
85 {
86  AVFilterContext *ctx = inlink->dst;
87  SCDetContext *s = ctx->priv;
89  int is_yuv = !(desc->flags & AV_PIX_FMT_FLAG_RGB) &&
90  (desc->flags & AV_PIX_FMT_FLAG_PLANAR) &&
91  desc->nb_components >= 3;
93  s->bitdepth = desc->comp[0].depth;
94  s->nb_planes = is_yuv ? 1 : av_pix_fmt_count_planes(inlink->format);
96  for (int plane = 0; plane < 4; plane++) {
97  ptrdiff_t line_size = av_image_get_linesize(inlink->format, inlink->w, plane);
98  s->width[plane] = line_size >> (s->bitdepth > 8);
99  s->height[plane] = inlink->h >> ((plane == 1 || plane == 2) ? desc->log2_chroma_h : 0);
100  }
102  s->sad = ff_scene_sad_get_fn(s->bitdepth == 8 ? 8 : 16);
103  if (!s->sad)
104  return AVERROR(EINVAL);
106  return 0;
107 }
110 {
111  SCDetContext *s = ctx->priv;
114 }
117 {
118  double ret = 0;
119  SCDetContext *s = ctx->priv;
122  if (prev_picref && frame->height == prev_picref->height
123  && frame->width == prev_picref->width) {
124  uint64_t sad = 0;
125  double mafd, diff;
126  uint64_t count = 0;
128  for (int plane = 0; plane < s->nb_planes; plane++) {
129  uint64_t plane_sad;
130  s->sad(prev_picref->data[plane], prev_picref->linesize[plane],
131  frame->data[plane], frame->linesize[plane],
132  s->width[plane], s->height[plane], &plane_sad);
133  sad += plane_sad;
134  count += s->width[plane] * s->height[plane];
135  }
137  emms_c();
138  mafd = (double)sad * 100. / count / (1ULL << s->bitdepth);
139  diff = fabs(mafd - s->prev_mafd);
140  ret = av_clipf(FFMIN(mafd, diff), 0, 100.);
141  s->prev_mafd = mafd;
142  av_frame_free(&prev_picref);
143  }
144  s->prev_picref = av_frame_clone(frame);
145  return ret;
146 }
148 static int set_meta(SCDetContext *s, AVFrame *frame, const char *key, const char *value)
149 {
150  return av_dict_set(&frame->metadata, key, value, 0);
151 }
154 {
155  int ret;
156  AVFilterLink *inlink = ctx->inputs[0];
157  AVFilterLink *outlink = ctx->outputs[0];
158  SCDetContext *s = ctx->priv;
159  AVFrame *frame;
161  FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink);
163  ret = ff_inlink_consume_frame(inlink, &frame);
164  if (ret < 0)
165  return ret;
167  if (frame) {
168  char buf[64];
169  s->scene_score = get_scene_score(ctx, frame);
170  snprintf(buf, sizeof(buf), "%0.3f", s->prev_mafd);
171  set_meta(s, frame, "lavfi.scd.mafd", buf);
172  snprintf(buf, sizeof(buf), "%0.3f", s->scene_score);
173  set_meta(s, frame, "lavfi.scd.score", buf);
175  if (s->scene_score > s->threshold) {
176  av_log(s, AV_LOG_INFO, "lavfi.scd.score: %.3f, lavfi.scd.time: %s\n",
177  s->scene_score, av_ts2timestr(frame->pts, &inlink->time_base));
178  set_meta(s, frame, "lavfi.scd.time",
179  av_ts2timestr(frame->pts, &inlink->time_base));
180  }
181  if (s->sc_pass) {
182  if (s->scene_score > s->threshold)
183  return ff_filter_frame(outlink, frame);
184  else {
185  av_frame_free(&frame);
186  }
187  } else
188  return ff_filter_frame(outlink, frame);
189  }
191  FF_FILTER_FORWARD_STATUS(inlink, outlink);
192  FF_FILTER_FORWARD_WANTED(outlink, inlink);
194  return FFERROR_NOT_READY;
195 }
197 static const AVFilterPad scdet_inputs[] = {
198  {
199  .name = "default",
200  .type = AVMEDIA_TYPE_VIDEO,
201  .config_props = config_input,
202  },
203  { NULL }
204 };
206 static const AVFilterPad scdet_outputs[] = {
207  {
208  .name = "default",
209  .type = AVMEDIA_TYPE_VIDEO,
210  },
211  { NULL }
212 };
215  .name = "scdet",
216  .description = NULL_IF_CONFIG_SMALL("Detect video scene change"),
217  .priv_size = sizeof(SCDetContext),
218  .priv_class = &scdet_class,
219  .uninit = uninit,
221  .inputs = scdet_inputs,
222  .outputs = scdet_outputs,
223  .activate = activate,
224 };
int ff_inlink_consume_frame(AVFilterLink *link, AVFrame **rframe)
Take a frame from the link&#39;s FIFO and update the link&#39;s stats.
Definition: avfilter.c:1491
#define NULL
Definition: coverity.c:32
void(* ff_scene_sad_fn)(SCENE_SAD_PARAMS)
Definition: scene_sad.h:34
int av_image_get_linesize(enum AVPixelFormat pix_fmt, int width, int plane)
Compute the size of an image line with format pix_fmt and width width for the plane plane...
Definition: imgutils.c:76
const AVPixFmtDescriptor * av_pix_fmt_desc_get(enum AVPixelFormat pix_fmt)
Definition: pixdesc.c:2573
This structure describes decoded (raw) audio or video data.
Definition: frame.h:314
Definition: opt.h:248
AVFilter ff_vf_scdet
Definition: vf_scdet.c:214
const char * desc
Definition: libsvtav1.c:79
planar YUV 4:4:4, 24bpp, (1 Cr & Cb sample per 1x1 Y samples)
Definition: pixfmt.h:71
misc image utilities
int av_pix_fmt_count_planes(enum AVPixelFormat pix_fmt)
Definition: pixdesc.c:2613
Main libavfilter public API header.
packed RGB 8:8:8, 24bpp, RGBRGB...
Definition: pixfmt.h:68
ff_scene_sad_fn ff_scene_sad_get_fn(int depth)
Definition: scene_sad.c:59
int bitdepth
Definition: vf_scdet.c:40
double scene_score
Definition: vf_scdet.c:43
AVFrame * prev_picref
Definition: vf_scdet.c:44
#define OFFSET(x)
Definition: vf_scdet.c:49
static const AVOption scdet_options[]
Definition: vf_scdet.c:53
static const AVFilterPad scdet_outputs[]
Definition: vf_scdet.c:206
#define AV_PIX_FMT_YUV420P12
Definition: pixfmt.h:403
const char * key
#define F
Definition: vf_scdet.c:51
AVFilterFormats * ff_make_format_list(const int *fmts)
Create a list of supported formats.
Definition: formats.c:287
const char * name
Pad name.
Definition: internal.h:60
static av_cold void uninit(AVFilterContext *ctx)
Definition: vf_scdet.c:109
AVFilterLink ** inputs
array of pointers to input links
Definition: avfilter.h:349
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
Definition: avfilter.c:1093
AVComponentDescriptor comp[4]
Parameters that describe how pixels are packed.
Definition: pixdesc.h:117
#define av_cold
Definition: attributes.h:88
timestamp utils, mostly useful for debugging/logging purposes
GLsizei GLboolean const GLfloat * value
Definition: opengl_enc.c:108
int64_t pts
Presentation timestamp in time_base units (time when frame should be shown to user).
Definition: frame.h:407
packed ABGR 8:8:8:8, 32bpp, ABGRABGR...
Definition: pixfmt.h:94
planar YUV 4:4:0 full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV440P and setting color_range...
Definition: pixfmt.h:100
planar YUV 4:2:2, 16bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV422P and setting col...
Definition: pixfmt.h:79
AVDictionary * metadata
Definition: frame.h:600
#define AV_PIX_FMT_YUV422P12
Definition: pixfmt.h:404
static int query_formats(AVFilterContext *ctx)
Definition: vf_scdet.c:63
#define av_log(a,...)
#define FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink)
Forward the status on an output link to an input link.
Definition: filters.h:199
ff_scene_sad_fn sad
Definition: vf_scdet.c:41
A filter pad used for either input or output.
Definition: internal.h:54
#define av_clipf
Definition: common.h:170
Scene SAD functions.
int width
Definition: frame.h:372
int sc_pass
Definition: vf_scdet.c:46
int ff_set_common_formats(AVFilterContext *ctx, AVFilterFormats *formats)
A helper for query_formats() which sets all links to the same list of formats.
Definition: formats.c:588
uint8_t log2_chroma_h
Amount to shift the luma height right to find the chroma height.
Definition: pixdesc.h:101
double threshold
Definition: vf_scdet.c:45
static __device__ float fabs(float a)
Definition: cuda_runtime.h:182
The pixel format contains RGB-like data (as opposed to YUV/grayscale).
Definition: pixdesc.h:148
void av_frame_free(AVFrame **frame)
Free the frame and any dynamically allocated objects in it, e.g.
Definition: frame.c:203
#define av_ts2timestr(ts, tb)
Convenience macro, the return value should be used only directly in function arguments but never stan...
Definition: timestamp.h:76
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:115
packed BGRA 8:8:8:8, 32bpp, BGRABGRA...
Definition: pixfmt.h:95
void * priv
private data for use by the filter
Definition: avfilter.h:356
simple assert() macros that are a bit more flexible than ISO C assert().
#define AV_PIX_FMT_YUV444P10
Definition: pixfmt.h:402
static int activate(AVFilterContext *ctx)
Definition: vf_scdet.c:153
GLsizei count
Definition: opengl_enc.c:108
packed RGBA 8:8:8:8, 32bpp, RGBARGBA...
Definition: pixfmt.h:93
planar YUV 4:2:2, 16bpp, (1 Cr & Cb sample per 2x1 Y samples)
Definition: pixfmt.h:70
uint64_t flags
Combination of AV_PIX_FMT_FLAG_...
Definition: pixdesc.h:106
#define AV_PIX_FMT_YUV422P9
Definition: pixfmt.h:397
static int set_meta(SCDetContext *s, AVFrame *frame, const char *key, const char *value)
Definition: vf_scdet.c:148
uint8_t nb_components
The number of components each pixel has, (1-4)
Definition: pixdesc.h:83
#define FFMIN(a, b)
Definition: common.h:105
ptrdiff_t height[4]
Definition: vf_scdet.c:38
planar YUV 4:2:0, 12bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV420P and setting col...
Definition: pixfmt.h:78
AVFormatContext * ctx
Definition: movenc.c:48
double prev_mafd
Definition: vf_scdet.c:42
these buffered frames must be flushed immediately if a new input produces new the filter must not call request_frame to get more It must just process the frame or queue it The task of requesting more frames is left to the filter s request_frame method or the application If a filter has several the filter must be ready for frames arriving randomly on any input any filter with several inputs will most likely require some kind of queuing mechanism It is perfectly acceptable to have a limited queue and to drop frames when the inputs are too unbalanced request_frame For filters that do not use the this method is called when a frame is wanted on an output For a it should directly call filter_frame on the corresponding output For a if there are queued frames already one of these frames should be pushed If the filter should request a frame on one of its repeatedly until at least one frame has been pushed Return or at least make progress towards producing a frame
#define s(width, name)
Definition: cbs_vp9.c:257
int nb_planes
Definition: vf_scdet.c:39
packed RGB 8:8:8, 24bpp, BGRBGR...
Definition: pixfmt.h:69
#define AV_PIX_FMT_YUV444P9
Definition: pixfmt.h:398
AVFrame * av_frame_clone(const AVFrame *src)
Create a new frame that references the same data as src.
Definition: frame.c:540
static const AVFilterPad outputs[]
Definition: af_acontrast.c:203
static int config_input(AVFilterLink *inlink)
Definition: vf_scdet.c:84
#define AV_LOG_INFO
Standard information.
Definition: log.h:205
int linesize[AV_NUM_DATA_POINTERS]
For video, size in bytes of each picture line.
Definition: frame.h:345
Descriptor that unambiguously describes how the bits of a pixel are stored in the up to 4 data planes...
Definition: pixdesc.h:81
these buffered frames must be flushed immediately if a new input produces new the filter must not call request_frame to get more It must just process the frame or queue it The task of requesting more frames is left to the filter s request_frame method or the application If a filter has several inputs
int av_dict_set(AVDictionary **pm, const char *key, const char *value, int flags)
Set the given entry in *pm, overwriting an existing entry.
Definition: dict.c:70
#define AV_PIX_FMT_YUV420P10
Definition: pixfmt.h:399
Describe the class of an AVClass context structure.
Definition: log.h:67
Filter definition.
Definition: avfilter.h:145
const char * name
Filter name.
Definition: avfilter.h:149
#define AV_PIX_FMT_YUV420P9
Definition: pixfmt.h:396
ptrdiff_t width[4]
Definition: vf_scdet.c:37
#define snprintf
Definition: snprintf.h:34
AVFilterLink ** outputs
array of pointers to output links
Definition: avfilter.h:353
static enum AVPixelFormat pix_fmts[]
Definition: libkvazaar.c:303
#define AV_PIX_FMT_YUV422P10
Definition: pixfmt.h:400
#define AV_PIX_FMT_YUV444P12
Definition: pixfmt.h:406
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
Definition: frame.h:328
The exact code depends on how similar the blocks are and how related they are to the and needs to apply these operations to the correct inlink or outlink if there are several Macros are available to factor that when no extra processing is inlink
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
Definition: pixfmt.h:66
Y , 8bpp.
Definition: pixfmt.h:74
planar YUV 4:4:4, 24bpp, full scale (JPEG), deprecated in favor of AV_PIX_FMT_YUV444P and setting col...
Definition: pixfmt.h:80
static double get_scene_score(AVFilterContext *ctx, AVFrame *frame)
Definition: vf_scdet.c:116
static av_always_inline int diff(const uint32_t a, const uint32_t b)
A list of supported formats for one end of a filter link.
Definition: formats.h:65
#define V
Definition: vf_scdet.c:50
static const AVFilterPad scdet_inputs[]
Definition: vf_scdet.c:197
An instance of a filter.
Definition: avfilter.h:341
int height
Definition: frame.h:372
planar YUV 4:4:0 (1 Cr & Cb sample per 1x2 Y samples)
Definition: pixfmt.h:99
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining all references to the list are updated That means that if a filter requires that its input and output have the same format amongst a supported all it has to do is use a reference to the same list of formats query_formats can leave some formats unset and return AVERROR(EAGAIN) to cause the negotiation mechanism toagain later.That can be used by filters with complex requirements to use the format negotiated on one link to set the formats supported on another.Frame references ownership and permissions
int depth
Number of bits in the component.
Definition: pixdesc.h:58
Pixel format.
Definition: pixfmt.h:64
At least one pixel component is not in the first data plane.
Definition: pixdesc.h:144